protortyp

Of course this comes from a European organisation.

No, the "action" part is the distinction. Their world model is conditioned on robot actions for example, which gives you two things the video gen alone can't: predict the future frames that follow a given action (change the action, get a different future from the same starting frame), and run it in reverse to infer the actions behind observed frames or output the actions needed to hit a goal (the output is motor commands abd not video frames).

I cut this on purpose due to the poor resolution other commenters have mentioned already.

Exactly. In an earlier draft I went deeper into this, but cut that section among others for brevity.

Turning my phone to grayscale solved the doomscrolling issue for me. Now I only browse HN all day long.

Fair, that's very helpful feedback.

> the interspersed animated/interactive models often don't seem strongly connected to the text

It's indeed the part I struggled with most. The intent was to make the constraint more "visceral", so that the "the interceptor can't catch up" point becomes something you feel by dragging a slider and wtaching the gap grow. But you're right that I didn't do enough to stitch each properly into the prose around it. It reads a bit too adjacent to the text.

For what it's worth, an earlier draft was nearly twice the length and even included a small missile-interception game as the introduction. I think cutting it was the right call though.

Thanks for the notes! I'll keep this in mind for the next post.

Curious which parts specifically felt that way for you? I spent over a week on this, and yes ofc, I used LLMs to help reformulate some sections, but "didn't seem to care enough to make it appealing to read" isn't it. Happy to look at the spots that felt choppy if you can point them out.

> I personally can't understand anyone wanting to move to the US anymore except for extreme reasons.

I am German and honestly can't wait to move to the US once I get a suitable H-1B offer. I already spent 8 months in Boston for a research stay, and back then the doomer mentality among natives was wild to me. From an outsider's perspective it’s crazy to watch. Life, and especially the ceiling for what you can achieve, is still 10x higher in the US than anywhere else.

I think people in the US severely underestimate how stagnant it feels in Europe and other continents right now. We are basically just stumbling from one crisis to the next, without any strong leadership (the US two-party system definitely has its advantages here, as you're able to charge fullspeed into one direction instead of not moving at all).

If you actually want to build ambitious things, the friction here is exhausting. And instead of being rewarded for high output you get taxed to death to prop up a system favored towards an aging/declining population. It's essentially a massive boomer tax. Younger workers have zero political leverage to change it because our demographic is just too small to matter at the ballot box.

Sure, the US definitely has its ugly sides, but if you want to work hard and actually capture the upside of what you build, it's still the only game in town. Even if that means jumping through all the hoops the current gov throws in your way.

I hope I can call myself an American one day.

What a time to be alive

Positives:

- Major breakthroughs in robotics research thanks to ultra scalable simulation software and RL

- AI will become even more useful in daily life

- New drug discoveries or successful stage 3 clinical studies related to / similar to GLP-1. Also in longevity research.

- Renewables will get even cheaper

- Creators will thrive with all the new AI tooling

- Government services will become more digital (at least in Germany)

- The war between Ukraine and Russia will end

- Interest rates continue to fall, causing an uptick in housing buildup

- The EU finally takes a harder stance against uncontrolled migration with concrete measures that will bring numbers far down.

Negatives:

- Societal loneliness will continue to accelerate

- People will use their brains even less, thanks to more widespread LLM use

- the Nazi party in Germany will become the majority in two states (but won't be able to govern them)

Completely unrelated, but I recently switched to using nix with home manager and nix-darwin to handle my entire dev setup and dotfiles.

If you work a lot in remote dev environments (I use coder a lot at work) that really does the trick.

I recently used Typst and their own collab solution for a paper we worked on. While some features are still lacking it was a pretty good experience overall.

I felt this on myself as well when I tried to use copilot. Especially when it was later in the day. I still use it for some boilerplate code / building visualizations that feels like boilerplate, but turned it off for any real important code. Atm I find most value in AI in discussing design decisions and to evaluate alternative approaches to mine. There it really had a huge positive impact on my workflow

This was already back in 2008/2009. At the time, I posted about my approach in online forums to ask for coding advice. Through those, I got to meet the creators of two more organizations that used a similar approach for PC gaming news pages.

And this was for the German market only. So I am quite sure that it was more common in the US at an earlier time, as they are usually ahead of us.

I doubt that the dogshit is so much worse than it used to be. Unfortunately, I also contributed to this. When I was 13 or 14, I scripted a little tool that allowed me to write one tech news blog post and generate multiple versions of it by simply swapping phrases, adding practically zero value. You'd be surprised (or not) how much of a tech news blog is repetitive wording, mainly to fill yet one more adsense banner after a new paragraph. It was not based on AI, but the core principle was the same as the example in the article.

I published each version on a separate WordPress blog covering roughly the same topic, chose random pictures, set random publish dates close to each other, and signed them all up for Google News. This non-AI dogshit dominated a small tech niche, making a decent amount of money at the time for a 14-year-old. I am pretty sure I was not the only one coming up with that idea at the time.

Can someone explain how Sam is able to run so many things at once? I am working on my startup and I barely find time to scroll HN during my toilet breaks.

I strongly agree. Much of my social circle went to Amsterdam in hopes of better salaries and the tax benefit for the first few years.

As a local, I'm really looking forward to when this madness is over in a few days, so we can finally use the metro again without that ever-present puke smell.

We also switched to Pulsar after running some benchmarks for our use cases. We use these services primarily as worker queues for image tasks that require low latency. And Pulsar turned out to have a 20x lower latency than Kafka in our setup.

Absolutely. I built my motorized one about 6 years ago (bought the motorized legs and wood top for a total of 450€ back then) and I still use it on all home office days. The key is having a rubber floor mat with some knobs you can play with to stand comfortably. I can also recommend getting a walking pad. I usually have 10k steps before lunch with a 2.5km/h speed only that lets me still type perfectly.

I primarily got my desk because I was tired of having bad posture and back pain. But since I also started working out heavily at the same time I built the table, I can't attribute the desk only to alleviating any pain I had. It definitely helps though with good posture.

I'm also happy I got a smart version that can remember 4 heights. I had another (much more expensive) table at my previous workplace that didn't have it. And it was a pain switching from a seated to standing position.

I already use gpt4 for this. Works quite well if you show it a couple of reference unit tests.

I agree. Even just a bunch of RTX 3060s go a long way[1]. Also, in the case of my startup we can use our local storage vs uploading dozens of terabytes of data to the cloud and deal with data privacy issues.

[1] https://timdettmers.com/2023/01/16/which-gpu-for-deep-learni...

The issue with lambda is that they're always booked out in my experience.

I primarily use folders and the dataview plugin [1] for 1). E.g. when I am managing a course, I have a structure like so:

https://www.dropbox.com/s/5mbcuu2pyy7eb3o/folder-structure.p...

I usually have a top-level note for a course, here the "Computational Surgineering.md". In there, I use the dataview plugin to simply create a dynamic table of all entries in the meetings subfolder:

https://www.dropbox.com/s/wk8jjldlohg6bf3/dataview.png?dl=0

Another option I use is nested tags[2], like #cs/meeting for the above use-case.

As for 2) I don't really use the global graph that much. It looks quite cool, but I primarily just look at a local graph with a maximum depth of 2-3 to quickly hop around.

[1] https://github.com/blacksmithgu/obsidian-dataview

[2] https://help.obsidian.md/Editing+and+formatting/Tags

Edit: Formatting is horrible on HN. I posted screenshots instead.

To me, Obsidian really became the killer app overall. I used Notion and Foam before and while the interface there is much more beautiful, I like having my files locally (synced via Syncthing) and being able to write my own plugins easily. The existing plugin ecosystem with instances like Dataview or Templater makes Obsidian such a great solution for personal and work planning, and as a knowledge base.

The only thing that's really missing is a collaborative version of Obsidian to be able to work in teams. I found craft.do that has a very similar feel, but it's quite pricey and, of course, you can't self-host.

I can also recommend using Syncthing with obsidian. I use it to synchronize my vault to all my devices. I also added it to my existing Raspberry Pi server that's always online, so I always have a distributor instance running and don't need to worry about sync issues.

Great write-up. I remember how shocked I was a few years back when we moved our backend to a micro-service architecture and our native Postgres installation to a docker container. As (I suppose) almost everyone, we also used ufw to manage the firewall.

Security should be the default stance, and any port exposed through docker should initially be restricted to local only. Going global should be explicit, imo.

Did flashing it with Linux have any impact on battery life? The main reason I switched from a Thinkpad T480 (2017) to a M2 Air two months ago was the horrible battery life of the Thinkpad. The M2 has been lasting me almost two days of work, compared to only 3 hours on the Thinkpad. But I miss quite a lot from linux

If you're interested in this topic, you might want to check out ASRG[1]. It's a community of safety and security professionals in the industry working to improve standards. They share their research openly and have regular in-person and online events where you can reach out to people.

[1] https://asrg.io/

I am quite happy that I am only using KeePassXC + Syncthing. My default assumption for these kinds of services is that they will be breached sooner or later.